Isotope-Labelling of Methyl Groups for NMR Studies of Large Proteins

MICHAEL J. PLEVIN AND JÉRÔME BOISBOUVIER

CEA, Institut de Biologie Structurale, CNRS, Institut de Biologie Structurale Jean-Pierre and Université Joseph Fourier, Institut de Biologie Structurale Jean-Pierre Ebel, Grenoble, France

E-mail: michael.plevin@ibs.fr or jerome.boisbouvier@ibs.fr

1.1 Introduction — Large Proteins and Solution NMR Spectroscopy

1.1.1 Isotope-Labelling and Protein NMR Spectroscopy

Solution NMR spectroscopy is a well-established technique for characterising the structure, function and dynamics of proteins at atomic resolution. Proteins are predominantly composed of carbon, nitrogen, oxygen and hydrogen. Of these four, only hydrogen has a naturally abundant, NMR-visible spin-½ nucleus and, for this reason, the proton was the major focus of early protein NMR studies. One of the major drawbacks of proton NMR spectroscopy is the inherent low dispersion of 1H chemical shifts. The narrow range of 1H resonance frequencies means that the ability to differentiate individual 1H signals becomes increasingly problematic as the size of the protein and therefore the number of potential signals increases.

The problem of low 1H signal overlap has now been largely overcome through the preparation of protein samples enriched with low natural abundance, spin-½ isotopes of carbon and/or nitrogen. Many NMR experiments have since been written that utilise the large signal dispersion of 13C or 15N nuclei to separate the signals of scalar-coupled nuclei over multiple dimensions. Furthermore, in addition to resolving spectral congestion, isotope enrichment introduces more NMR-visible probes into the molecules of interest and allows a multitude of structural and dynamic information to be accessed from their NMR signals.

Isotopic enrichment of proteins can take two forms: uniform or selective. In the most commonly used approach the recombinant target protein is over-expressed from E. coli grown in an isotopically enriched minimal-expression medium containing uniformly labelled [13C]glucose and/or [15N]ammonium chloride or sulphate, as the only carbon and nitrogen sources. The resulting protein product is isotopically enriched at the same level as the expression medium. Uniform labelling approaches were developed towards the end of the 1980s (ref. 4) and since have become routine and robust. In the last 20 years, the price of isotopically enriched reagents has decreased considerably making uniform labelling a common practice in structural biology laboratories.

Isotope-labelling of individual amino acids or groups of amino acids can also be performed. Residue-specific isotope labelling is achieved by supplementing the expression medium with isotopically enriched amino acids. This approach is somewhat limited in vivo as a result of the scrambling of the isotope-labelled sites by bacterial metabolic pathways. As an alternative, isotope-labelled amino acids can be used in combination with cell-free in vitro expression systems, which essentially alleviate isotopic dilution.

1.1.2 General Considerations for NMR Studies of Larger Proteins

Over the past 20 years an enormous array of multi-dimensional heteronuclear NMR experiments have been designed that can extract structural or dynamic information about isotopically enriched proteins. The strategy of combining isotope-labelling with tailored NMR experiments has been so successful that it has encouraged NMR spectroscopists to study larger and more complicated biomolecular systems. However, as the size of protein targets increases new problems arise.

The lifetime of the excited state in NMR spectroscopy is predominantly affected by the overall molecular tumbling rate. As molecular size increases the tumbling rate slows and this leads to an increase in the rate at which transverse magnetisation relaxes. As the linewidth of an NMR signal is proportional to the transverse relaxation rate, NMR spectra of larger molecules which tumble more slowly are characterised by broad NMR signals.

The short lifetime of transverse relaxation in large proteins severely affects the sensitivity, effectiveness and scope of NMR experiments. NMR pulse sequences frequently rely on scalar couplings to transfer magnetisation between nuclei of interest. Such transfer steps require periods in which nuclear magnetisation is the transverse plane and therefore subject to transverse relaxation. Thus, complicated pulse sequences that correlate nuclei via weak scalar couplings or that require multiple transfers mediated by scalar couplings become less effective and less sensitive for larger proteins.

Resonance assignment of proton, carbon and nitrogen nuclei in the polypeptide backbone is a critical first step in many NMR studies of protein structure, dynamics or interactions. A common starting point is a two-dimensional (2D) (1H,15N) heteronuclear correlation spectrum acquired, for example, using the Heteronuclear Single Quantum Coherence (HSQC) experiment. An in-depth assessment of NMR data requires being able to locate each NH cross-peak to a unique site in the target protein. This is achieved by determining sequence-specific resonance assignments. There are numerous experimental strategies that facilitate backbone resonance assignment, many of which make use of uniform isotope-labelling strategies and multi-dimensional heteronuclear NMR experiments. While these approaches work well for smaller proteins (<25 kDa; Figure 1.1), they cease to be applicable when the molecular weight increases as the transverse magnetisation relaxes more rapidly.

The major source of relaxation for 1H nuclei in higher molecular weight proteins is the large number of dipolar...